Learning to use an invisible visual signal for perception

How does the brain construct a percept from sensory signals? One approach to this fundamental question is to investigate perceptual learning as induced by exposure to statistical regularities in sensory signals [1-7]. Recent studies showed that exposure to novel correlations between sensory signals can cause a signal to have new perceptual effects [2, 3]. In those studies, however, the signals were clearly visible. The automaticity of the learning was therefore difficult to determine. Here we investigate whether learning of this sort, which causes new effects on appearance, can be low level and automatic by employing a visual signal whose perceptual consequences were made invisible-a vertical disparity gradient masked by other depth cues. This approach excluded high-level influences such as attention or consciousness. Our stimulus for probing perceptual appearance was a rotating cylinder. During exposure, we introduced a new contingency between the invisible signal and the rotation direction of the cylinder. When subsequently presenting an ambiguously rotating version of the cylinder, we found that the invisible signal influenced the perceived rotation direction. This demonstrates that perception can rapidly undergo "structure learning" by automatically picking up novel contingencies between sensory signals, thus automatically recruiting signals for novel uses during the construction of a percept.     

Adaptation to visual feedback delay influences visuomotor learning

Computational theory of motor control suggests that the brain continuously monitors motor commands, to predict their sensory consequences before actual sensory feedback becomes available. Such prediction error is a driving force of motor learning, and therefore appropriate associations between motor commands and delayed sensory feedback signals are crucial. Indeed, artificially introduced delays in visual feedback have been reported to degrade motor learning. However, considering our perceptual ability to causally bind our own actions with sensory feedback, demonstrated by the decrease in the perceived time delay following repeated exposure to an artificial delay, we hypothesized that such perceptual binding might alleviate deficits of motor learning associated with delayed visual feedback. Here, we evaluated this hypothesis by investigating the ability of human participants to adapt their reaching movements in response to a novel visuomotor environment with 3 visual feedback conditions--no-delay, sudden-delay, and adapted-delay. To introduce novelty into the trials, the cursor position, which originally indicated the hand position in baseline trials, was rotated around the starting position. In contrast to the no-delay condition, a 200-ms delay was artificially introduced between the cursor and hand positions during the presence of visual rotation (sudden-delay condition), or before the application of visual rotation (adapted-delay condition). We compared the learning rate (representing how the movement error modifies the movement direction in the subsequent trial) between the 3 conditions. In comparison with the no-delay condition, the learning rate was significantly degraded for the sudden-delay condition. However, this degradation was significantly alleviated by prior exposure to the delay (adapted-delay condition). Our data indicate the importance of appropriate temporal associations between motor commands and sensory feedback in visuomotor learning. Moreover, they suggest that the brain is able to account for such temporal associations in a flexible manner.     

Reinforcement Learning on Slow Features of High-Dimensional Input Streams

Humans and animals are able to learn complex behaviors based on a massive stream of sensory information from different modalities. Early animal studies have identified learning mechanisms that are based on reward and punishment such that animals tend to avoid actions that lead to punishment whereas rewarded actions are reinforced. However, most algorithms for reward-based learning are only applicable if the dimensionality of the state-space is sufficiently small or its structure is sufficiently simple. Therefore, the question arises how the problem of learning on high-dimensional data is solved in the brain. In this article, we propose a biologically plausible generic two-stage learning system that can directly be applied to raw high-dimensional input streams. The system is composed of a hierarchical slow feature analysis (SFA) network for preprocessing and a simple neural network on top that is trained based on rewards. We demonstrate by computer simulations that this generic architecture is able to learn quite demanding reinforcement learning tasks on high-dimensional visual input streams in a time that is comparable to the time needed when an explicit highly informative low-dimensional state-space representation is given instead of the high-dimensional visual input. The learning speed of the proposed architecture in a task similar to the Morris water maze task is comparable to that found in experimental studies with rats. This study thus supports the hypothesis that slowness learning is one important unsupervised learning principle utilized in the brain to form efficient state representations for behavioral learning.     

A theory of cortical responses

This article concerns the nature of evoked brain responses and the principles underlying their generation. We start with the premise that the sensory brain has evolved to represent or infer the causes of changes in its sensory inputs. The problem of inference is well formulated in statistical terms. The statistical fundaments of inference may therefore afford important constraints on neuronal implementation. By formulating the original ideas of Helmholtz on perception, in terms of modern-day statistical theories, one arrives at a model of perceptual inference and learning that can explain a remarkable range of neurobiological facts.It turns out that the problems of inferring the causes of sensory input (perceptual inference) and learning the relationship between input and cause (perceptual learning) can be resolved using exactly the same principle. Specifically, both inference and learning rest on minimizing the brain's free energy, as defined in statistical physics. Furthermore, inference and learning can proceed in a biologically plausible fashion. Cortical responses can be seen as the brain's attempt to minimize the free energy induced by a stimulus and thereby encode the most likely cause of that stimulus. Similarly, learning emerges from changes in synaptic efficacy that minimize the free energy, averaged over all stimuli encountered. The underlying scheme rests on empirical Bayes and hierarchical models of how sensory input is caused. The use of hierarchical models enables the brain to construct prior expectations in a dynamic and context-sensitive fashion. This scheme provides a principled way to understand many aspects of cortical organization and responses. The aim of this article is to encompass many apparently unrelated anatomical, physiological and psychophysical attributes of the brain within a single theoretical perspective.In terms of cortical architectures, the theoretical treatment predicts that sensory cortex should be arranged hierarchically, that connections should be reciprocal and that forward and backward connections should show a functional asymmetry (forward connections are driving, whereas backward connections are both driving and modulatory). In terms of synaptic physiology, it predicts associative plasticity and, for dynamic models, spike-timing-dependent plasticity. In terms of electrophysiology, it accounts for classical and extra classical receptive field effects and long-latency or endogenous components of evoked cortical responses. It predicts the attenuation of responses encoding prediction error with perceptual learning and explains many phenomena such as repetition suppression, mismatch negativity (MMN) and the P300 in electroencephalography. In psychophysical terms, it accounts for the behavioural correlates of these physiological phenomena, for example, priming and global precedence. The final focus of this article is on perceptual learning as measured with the MMN and the implications for empirical studies of coupling among cortical areas using evoked sensory responses.     

Structure Learning in a Sensorimotor Association Task

Learning is often understood as an organism''s gradual acquisition of the association between a given sensory stimulus and the correct motor response. Mathematically, this corresponds to regressing a mapping between the set of observations and the set of actions. Recently, however, it has been shown both in cognitive and motor neuroscience that humans are not only able to learn particular stimulus-response mappings, but are also able to extract abstract structural invariants that facilitate generalization to novel tasks. Here we show how such structure learning can enhance facilitation in a sensorimotor association task performed by human subjects. Using regression and reinforcement learning models we show that the observed facilitation cannot be explained by these basic models of learning stimulus-response associations. We show, however, that the observed data can be explained by a hierarchical Bayesian model that performs structure learning. In line with previous results from cognitive tasks, this suggests that hierarchical Bayesian inference might provide a common framework to explain both the learning of specific stimulus-response associations and the learning of abstract structures that are shared by different task environments.     

Shaping embodied neural networks for adaptive goal-directed behavior

The acts of learning and memory are thought to emerge from the modifications of synaptic connections between neurons, as guided by sensory feedback during behavior. However, much is unknown about how such synaptic processes can sculpt and are sculpted by neuronal population dynamics and an interaction with the environment. Here, we embodied a simulated network, inspired by dissociated cortical neuronal cultures, with an artificial animal (an animat) through a sensory-motor loop consisting of structured stimuli, detailed activity metrics incorporating spatial information, and an adaptive training algorithm that takes advantage of spike timing dependent plasticity. By using our design, we demonstrated that the network was capable of learning associations between multiple sensory inputs and motor outputs, and the animat was able to adapt to a new sensory mapping to restore its goal behavior: move toward and stay within a user-defined area. We further showed that successful learning required proper selections of stimuli to encode sensory inputs and a variety of training stimuli with adaptive selection contingent on the animat's behavior. We also found that an individual network had the flexibility to achieve different multi-task goals, and the same goal behavior could be exhibited with different sets of network synaptic strengths. While lacking the characteristic layered structure of in vivo cortical tissue, the biologically inspired simulated networks could tune their activity in behaviorally relevant manners, demonstrating that leaky integrate-and-fire neural networks have an innate ability to process information. This closed-loop hybrid system is a useful tool to study the network properties intermediating synaptic plasticity and behavioral adaptation. The training algorithm provides a stepping stone towards designing future control systems, whether with artificial neural networks or biological animats themselves.     

Isotropic sequence order learning using a novel linear algorithm in a closed loop behavioural system

In this article, we present an isotropic algorithm for sequence order learning. Its central goal is to learn the causal relation between two (or more) inputs in order to react to the earliest incoming signal after successful learning (like in typical classical conditioning situations). We implement this algorithm in a behaving system (a robot) thereby creating a closed loop situation where the learner's actions influence its own sensor inputs to the end of creating an autonomous agent. Autonomous behaviour implies that learning goals are internally defined within the organism's capabilities. Standard learning models for sequence learning (e.g. temporal difference (TD)-learning) need an externally defined reward. This, however, is in conflict with the requirement of an implicitly defined internal goal in autonomous behaviour. Therefore, in this study we present a system in which the external reward is replaced by a reflex loop. This loop explicitly includes the environment. Every reflex loop has the inherent disadvantage, which is that its re-actions occur each time just after a reflex-eliciting sensor event and thus 'too late'. However, a reflex can serve as the internal reference for sequence order learning, which has the task of eliminating this disadvantage by creating earlier anticipatory actions. In our system learning is achieved by modifying synaptic weights of a linear neuron with a correlation based learning rule which involves the derivative of the neuron's output. All input lines are entirely isotropic. The synaptic weight change curve of this rule is strongly related to the temporal Hebb learning rule, which was found in spike timing experiments. We find that after learning the reflex loop is replaced in functional terms with an earlier anticipatory action (and pathway). In addition, we observed that the synaptic weights stabilise as soon as the reflex remains silent.     

Learning cortical topography from spatiotemporal stimuli

Stimulus representation is a functional interpretation of early sensory cortices. Early sensory cortices are subject to stimulus-induced modifications. Common models for stimulus-induced learning within topographic representations are based on the stimuli's spatial structure and probability distribution. Furthermore, we argue that average temporal stimulus distances reflect the stimuli's relatedness. As topographic representations reflect the stimuli's relatedness, the temporal structure of incoming stimuli is important for the learning in cortical maps. Motivated by recent neurobiological findings, we present an approach of cortical self-organization that additionally takes temporal stimulus aspects into account. The proposed model transforms average interstimulus intervals into representational distances. Thereby, neural topography is related to stimulus dynamics. This offers a new time-based interpretation of cortical maps. Our approach is based on a wave-like spread of cortical activity. Interactions between dynamics and feedforward activations lead to shifts of neural activity. The psychophysical saltation phenomenon may represent an analogue to the shifts proposed here. With regard to cortical plasticity, we offer an explanation for neurobiological findings that other models cannot explain. Moreover, we predict cortical reorganizations under new experimental, spatiotemporal conditions. With regard to psychophysics, we relate the saltation phenomenon to dynamics and interaction in early sensory cortices and predict further effects in the perception of spatiotemporal stimuli. Received: 17 March 1999 / Accepted in revised form: 10 August 1999     

Prediction precedes control in motor learning

Skilled motor behavior relies on the brain learning both to control the body and predict the consequences of this control. Prediction turns motor commands into expected sensory consequences, whereas control turns desired consequences into motor commands. To capture this symmetry, the neural processes underlying prediction and control are termed the forward and inverse internal models, respectively. Here, we investigate how these two fundamental processes are related during motor learning. We used an object manipulation task in which subjects learned to move a hand-held object with novel dynamic properties along a prescribed path. We independently and simultaneously measured subjects' ability to control their actions and to predict their consequences. We found different time courses for predictor and controller learning, with prediction being learned far more rapidly than control. In early stages of manipulating the object, subjects could predict the consequences of their actions, as measured by the grip force they used to grasp the object, but could not generate appropriate actions for control, as measured by their hand trajectory. As predicted by several recent theoretical models of sensorimotor control, our results indicate that people can learn to predict the consequences of their actions before they can learn to control their actions.     

In vivo induction of oocyte maturation and ovulation in zebrafish

The maturation of fish oocytes is a well-characterized system induced by progestins via non-genomic actions. In a previous study, we demonstrated that diethylstilbestrol (DES), a non-steroidal estrogen, induces fish oocyte maturation via the membrane progestin receptor (mPR). Here, we attempted to evaluate the effect of DES as an environmental endocrine disrupting chemical (EDC) upon fish oocyte maturation using live zebrafish. DES triggered oocyte maturation within several hours in vivo when administrated directly into the surrounding water. The natural teleost maturation-inducing hormone, 17alpha, 20beta-dihydroxy-4-pregnen-3-one (17,20beta-DHP) also induced oocyte maturation in vivo. Steroids such as testosterone, progesterone or 17alpha-hydroxyprogesterone were also effective in vivo. Further studies indicated that externally applied 17,20beta-DHP even induced ovulation. In contrast to 17,20beta -DHP, DES induced maturation but not ovulation. Theoretically this assay system provides a means to distinguish pathways involved in the induction of ovulation, which are known to be induced by genomic actions from the pathway normally involved in the induction of oocyte maturation, a typical non-genomic action-dependent pathway. In summary, we have demonstrated the effect of EDCs on fish oocyte maturation in vivo. To address the effects, we have explored a conceptually new approach to distinguish between the genomic and non-genomic actions induced by steroids. The assay can be applied to screens of progestin-like effects upon oocyte maturation and ovulation for small molecules of pharmacological agents or EDCs.     

Brain–computer interfaces for dissecting cognitive processes underlying sensorimotor control

Sensorimotor control engages cognitive processes such as prediction, learning, and multisensory integration. Understanding the neural mechanisms underlying these cognitive processes with arm reaching is challenging because we currently record only a fraction of the relevant neurons, the arm has nonlinear dynamics, and multiple modalities of sensory feedback contribute to control. A brain–computer interface (BCI) is a well-defined sensorimotor loop with key simplifying advantages that address each of these challenges, while engaging similar cognitive processes. As a result, BCI is becoming recognized as a powerful tool for basic scientific studies of sensorimotor control. Here, we describe the benefits of BCI for basic scientific inquiries and review recent BCI studies that have uncovered new insights into the neural mechanisms underlying sensorimotor control.     

The Brain Ages Optimally to Model Its Environment: Evidence from Sensory Learning over the Adult Lifespan

The aging brain shows a progressive loss of neuropil, which is accompanied by subtle changes in neuronal plasticity, sensory learning and memory. Neurophysiologically, aging attenuates evoked responses—including the mismatch negativity (MMN). This is accompanied by a shift in cortical responsivity from sensory (posterior) regions to executive (anterior) regions, which has been interpreted as a compensatory response for cognitive decline. Theoretical neurobiology offers a simpler explanation for all of these effects—from a Bayesian perspective, as the brain is progressively optimized to model its world, its complexity will decrease. A corollary of this complexity reduction is an attenuation of Bayesian updating or sensory learning. Here we confirmed this hypothesis using magnetoencephalographic recordings of the mismatch negativity elicited in a large cohort of human subjects, in their third to ninth decade. Employing dynamic causal modeling to assay the synaptic mechanisms underlying these non-invasive recordings, we found a selective age-related attenuation of synaptic connectivity changes that underpin rapid sensory learning. In contrast, baseline synaptic connectivity strengths were consistently strong over the decades. Our findings suggest that the lifetime accrual of sensory experience optimizes functional brain architectures to enable efficient and generalizable predictions of the world.     

The cerebellum updates predictions about the visual consequences of one's behavior

Each action has sensory consequences that need to be distinguished from sensations arising from the environment. This is accomplished by the comparing of internal predictions about these consequences with the actual afference, thereby isolating the afferent component that is self-produced. Because the sensory consequences of actions vary as a result of changes of the effector's efficacy, internal predictions need to be updated continuously and on a short time scale. Here, we tested the hypothesis that this updating of predictions about the sensory consequences of actions is mediated by the cerebellum, a notion that parallels the cerebellum's role in motor learning. Patients with cerebellar lesions and their matched controls were equally able to detect experimental modifications of visual feedback about their pointing movements. When such feedback was constantly rotated, both groups instantly attributed the visual feedback to their own actions. However, in interleaved trials without actual feedback, patients did no longer account for this feedback rotation--neither perceptually nor with respect to motor performance. Both deficits can be explained by an impaired updating of internal predictions about the sensory consequences of actions caused by cerebellar pathology. Thus, the cerebellum guarantees both precise performance and veridical perceptual interpretation of actions.     

Mental imaging of motor activity in humans

Motor imagery corresponds to a subliminal activation of the motor system, a system that appears to be involved not only in producing movements, but also in imagining actions, recognising tools and learning by observation, as well as in understanding the behaviour of other people. Recent advances in the field include the use of techniques for mapping brain activity and probing cortical excitability, as well as observation of brain lesioned patients during imaging tasks; these advances provide new insights into the covert aspects of motor activity.     

Task-induced neural covariability as a signature of approximate Bayesian learning and inference

Perception is often characterized computationally as an inference process in which uncertain or ambiguous sensory inputs are combined with prior expectations. Although behavioral studies have shown that observers can change their prior expectations in the context of a task, robust neural signatures of task-specific priors have been elusive. Here, we analytically derive such signatures under the general assumption that the responses of sensory neurons encode posterior beliefs that combine sensory inputs with task-specific expectations. Specifically, we derive predictions for the task-dependence of correlated neural variability and decision-related signals in sensory neurons. The qualitative aspects of our results are parameter-free and specific to the statistics of each task. The predictions for correlated variability also differ from predictions of classic feedforward models of sensory processing and are therefore a strong test of theories of hierarchical Bayesian inference in the brain. Importantly, we find that Bayesian learning predicts an increase in so-called “differential correlations” as the observer’s internal model learns the stimulus distribution, and the observer’s behavioral performance improves. This stands in contrast to classic feedforward encoding/decoding models of sensory processing, since such correlations are fundamentally information-limiting. We find support for our predictions in data from existing neurophysiological studies across a variety of tasks and brain areas. Finally, we show in simulation how measurements of sensory neural responses can reveal information about a subject’s internal beliefs about the task. Taken together, our results reinterpret task-dependent sources of neural covariability as signatures of Bayesian inference and provide new insights into their cause and their function.     

Molecular mapping of movement-associated areas in the avian brain: a motor theory for vocal learning origin

Vocal learning is a critical behavioral substrate for spoken human language. It is a rare trait found in three distantly related groups of birds-songbirds, hummingbirds, and parrots. These avian groups have remarkably similar systems of cerebral vocal nuclei for the control of learned vocalizations that are not found in their more closely related vocal non-learning relatives. These findings led to the hypothesis that brain pathways for vocal learning in different groups evolved independently from a common ancestor but under pre-existing constraints. Here, we suggest one constraint, a pre-existing system for movement control. Using behavioral molecular mapping, we discovered that in songbirds, parrots, and hummingbirds, all cerebral vocal learning nuclei are adjacent to discrete brain areas active during limb and body movements. Similar to the relationships between vocal nuclei activation and singing, activation in the adjacent areas correlated with the amount of movement performed and was independent of auditory and visual input. These same movement-associated brain areas were also present in female songbirds that do not learn vocalizations and have atrophied cerebral vocal nuclei, and in ring doves that are vocal non-learners and do not have cerebral vocal nuclei. A compilation of previous neural tracing experiments in songbirds suggests that the movement-associated areas are connected in a network that is in parallel with the adjacent vocal learning system. This study is the first global mapping that we are aware for movement-associated areas of the avian cerebrum and it indicates that brain systems that control vocal learning in distantly related birds are directly adjacent to brain systems involved in movement control. Based upon these findings, we propose a motor theory for the origin of vocal learning, this being that the brain areas specialized for vocal learning in vocal learners evolved as a specialization of a pre-existing motor pathway that controls movement.     

An Adapting Auditory-motor Feedback Loop Can Contribute to Generating Vocal Repetition

Consecutive repetition of actions is common in behavioral sequences. Although integration of sensory feedback with internal motor programs is important for sequence generation, if and how feedback contributes to repetitive actions is poorly understood. Here we study how auditory feedback contributes to generating repetitive syllable sequences in songbirds. We propose that auditory signals provide positive feedback to ongoing motor commands, but this influence decays as feedback weakens from response adaptation during syllable repetitions. Computational models show that this mechanism explains repeat distributions observed in Bengalese finch song. We experimentally confirmed two predictions of this mechanism in Bengalese finches: removal of auditory feedback by deafening reduces syllable repetitions; and neural responses to auditory playback of repeated syllable sequences gradually adapt in sensory-motor nucleus HVC. Together, our results implicate a positive auditory-feedback loop with adaptation in generating repetitive vocalizations, and suggest sensory adaptation is important for feedback control of motor sequences.     

Neurotoxic Effects of Methamphetamine on Rat Hippocampus Pyramidal Neurons

Methamphetamine (MAP) is known to alter behavior and cause deficits in learning and memory. While the major site of action of MAP is on mesolimbic dopaminergic pathways, the effects on learning and memory raise the possibility of important actions in the hippocampus. We have studied electrophysiologic and morphologic effects of MAP in the CA1 region of hippocampus from young male rats chronically exposed to MAP, male rats exposed during gestation only and the effects of bath perfusion of MAP onto brain slices from control rats. Pyramidal neurons in brain slices from chronically exposed rats had reduced membrane potential and membrane resistance. Long-term potentiation (LTP) was reduced as compared to control, but when MAP was acutely perfused over control slices the amplitude of LTP was increased. LTP in young adult animals that had been gestationally exposed to MAP showed reduced LTP as compared to controls. Morphologically CA1 pyramidal neurons in chronically exposed animals showed a high prevalence of extensive blebbing of dendrites. We conclude that the NMDA receptor and the process of LTP are also targets of MAP dysfunction, at least in the hippocampus.     

Attentional modulation strength in cortical area MT depends on stimulus contrast

The attentional modulation of sensory information processing in the visual system is the result of top-down influences, which can cause a multiplicative modulation of the firing rate of sensory neurons in extrastriate visual cortex, an effect reminiscent of the bottom-up effect of changes in stimulus contrast. This similarity could simply reflect the multiplicity of both effects. But, here we show that in direction-selective neurons in monkey visual cortical area MT, stimulus and attentional effects share a nonlinearity. These neurons show higher response gain for both contrast and attentional changes for intermediate contrast stimuli and smaller gain for low- and high-contrast stimuli. This finding suggests a close relationship between the neural encoding of stimulus contrast and the modulating effect of the behavioral relevance of stimuli.     

Testing simulation theory with cross-modal multivariate classification of fMRI data

The discovery of mirror neurons has suggested a potential neural basis for simulation and common coding theories of action perception, theories which propose that we understand other people's actions because perceiving their actions activates some of our neurons in much the same way as when we perform the actions. We propose testing this model directly in humans with functional magnetic resonance imaging (fMRI) by means of cross-modal classification. Cross-modal classification evaluates whether a classifier that has learned to separate stimuli in the sensory domain can also separate the stimuli in the motor domain. Successful classification provides support for simulation theories because it means that the fMRI signal, and presumably brain activity, is similar when perceiving and performing actions. In this paper we demonstrate the feasibility of the technique by showing that classifiers which have learned to discriminate whether a participant heard a hand or a mouth action, based on the activity patterns in the premotor cortex, can also determine, without additional training, whether the participant executed a hand or mouth action. This provides direct evidence that, while perceiving others' actions, (1) the pattern of activity in premotor voxels with sensory properties is a significant source of information regarding the nature of these actions, and (2) that this information shares a common code with motor execution.